Survey Design

What Is Likert Scale?

An ordered rating scale — commonly 5 or 7 points, from "Strongly Disagree" to "Strongly Agree" — used to measure agreement, satisfaction, or frequency.

A Likert scale presents a statement (“The onboarding process was easy to follow”) with a range of ordered response options — typically five (Strongly Disagree, Disagree, Neutral, Agree, Strongly Agree) or seven for finer granularity. It’s named after psychologist Rensis Likert, who introduced the method in 1932.

Strictly speaking, a “Likert scale” is a set of several such statements summed into one score; a single rating item is technically a “Likert-type item.” In practice, both are called Likert scales, and most surveys use the single-item version.

One real design decision: odd-numbered scales (5 or 7 points) include a neutral midpoint, which lets genuinely undecided respondents answer honestly. Even-numbered scales (4 or 6 points) force a lean one way or the other — useful when you specifically don’t want a “safe” neutral answer, but it can also push uncertain respondents into a false positive or negative.

Example

A 5-point Likert item
The checkout process was easy to complete.
  • Strongly Disagree
  • Disagree
  • Neutral
  • Agree
  • Strongly Agree

What it does not tell you

  • The answers are ordered categories. The distance from "Agree" to "Strongly agree" is not necessarily the same as from "Neutral" to "Agree", so an average is a convenience, and the share choosing the top options is often easier to defend.
  • Wording and the labels on each point change the results. Keep both fixed when you compare one round with another.
  • People differ in how willing they are to choose the end points, which can distort comparisons between groups.

Sources

  • Likert, R. (1932). A technique for the measurement of attitudes. Archives of Psychology, 22(140), 1-55.

How to Use Likert Scales in SurveyRock

Build Likert questions as matrix/grid questions in SurveyRock to ask several statements on the same scale in one compact table, or as standalone rating questions for a single statement. Response dashboards break out the distribution across each option automatically.

Try the survey builder